Rainfall- Runoff Modelling using HEC-HMS and ANN for Shiavde, Upper Krishna Basin, India

Abstract

Accurate estimation of runoff at a specific location in a basin helps in optimizing different water resource systems. This process is complex and nonlinear in nature owing mostly to the random nature of its most important parameter that is ‘rainfall’. In the present paper rainfall runoff modelling is attempted to forecast one day ahead runoff for the Shivade basin in the upper Krishna basin of Maharashtra, India using Artificial Neural Networks (ANNs) and numerical model-Hydrologic Engineering Centre-Hydrologic Modelling System (HEC-HMS). In the present exercise three independent years were used for validation purpose. Total 17 years of daily rainfall and discharge data from 1999 to 2013 was considered for model development exercise. Performance of the models quantitively is evaluated via different statistical indicators like correlation coefficient (r), root mean square error (RMSE) and coefficient of efficiency (CE). Additionally, qualitatively evaluation also done by drawing hydrographs and scatter plots. ANN model performance is better compared to HEC-HMS model with higher correlation coefficient (0.86) and lower RMSE (103.71 m3/s). The present attempt is unique because it uses restricted basin data to develop models. However, both models performed poorly at forecasting extreme events. It can be said that, the ANN model in conjunction with the HEC-HMS could be employed as an additional strategy (if not alternative) to address rainfall-runoff process.Keywords: Daily Rainfall -Runoff model; HEC-HMS; ANN; Average Mutual Information (AMI); QGIS software.

Keywords

• keywords: daily rainfall-runoff model
• hec-hms
• ann
• average mutual information (ami)
• qgis software